Pending Pod
- Docker & Kubernetes course
- Intermediate
- Lesson 76/82
- 4 min
Learn to inspect Pending Pods.
A pending pod
A pod stuck in Pending shows that the scheduler couldn't place it on any node (or that the image is being pulled). The FailedScheduling message in the event tells the reason:
0/3 nodes are available:
3 Insufficient cpu. # requests too big
1 node(s) had untolerated taint {node-role...} # taint
2 node(s) didn't match Pod's node affinity/selector # nodeSelector
pod has unbound immediate PersistentVolumeClaims # no storageInspecting node capacity
kubectl describe node shows the allocatable resources and the total requests of the pods on the node as percentages. If the resources the pod requests exceed the free capacity, either the request is reduced or a node is added:
$ kubectl describe node worker-1 | grep -A6 "Allocated resources"
Allocated resources:
Resource Requests Limits
cpu 1850m (92%) 3 (150%)
memory 3200Mi (81%) 4Gi (104%)
$ kubectl get pod api-xyz -o jsonpath='{.spec.containers[0].resources.requests}'
{"cpu":"500m","memory":"1Gi"}Practice Why isn't it placed?
Write three commands: list the pods in the Pending phase; print the requests of the api-xyz pod with jsonpath='{.spec.containers[0].resources.requests}'; show the worker-1 node in detail.
Starter code
# buraya yazSample solution
kubectl get pods --field-selector status.phase=Pending
kubectl get pod api-xyz -o jsonpath='{.spec.containers[0].resources.requests}'
kubectl describe node worker-1Expected output
NAME READY STATUS RESTARTS AGE
api-xyz 0/1 Pending 0 4m
{"cpu":"500m","memory":"1Gi"}
Name: worker-1
Allocated resources:
memory 3200Mi (81%) 4Gi (104%)
Questions and answers
Review the lesson questions together with their answers. To test yourself, solve the lesson interactively.
1 What is the right first step for the "3 Insufficient memory" message?
- Changing the namespace
- Comparing the pod's requests.memory with the nodes' allocatable memory Correct answer
- Deleting and recreating the pod
- Shrinking the image
Explanation: The scheduler looks at requests, not usage.
2 List only the Pending pods.
kubectl get pods --field-selector status.phase=____kubectl get pods --field-selector status.phase=PendingExplanation: status.phase is the pod phase.
3 Match the messages.
| Term | Meaning |
|---|---|
Insufficient cpu |
Request exceeds capacity |
untolerated taint |
The node repels it |
node affinity/selector |
No node with the right label |
unbound PersistentVolumeClaims |
Storage not bound |